reconsider the uniqueness in go n steps [bidirectly]

[whitewum]

Consider a very simple example:
who is my friend of friend (fof), or who is my friend of a friend of friend (fofof...)?

By the first though, we could use go 2 steps from $me over friendship_edge.
But wait, does the sentence return and only return fof?
Q1. should the starting node (me) a fof?
Q2. can a (both 1-hop & 2-hop) friend be a fof too?

As the cypher spec said: "While pattern matching, Cypher makes sure to not include matches where the same graph relationship is found multiple times in a single pattern. ".

MATCH (user:User {name: 'Adam'})-[r1:FRIEND]-()-[r2:FRIEND]-(friend_of_a_friend) RETURN friend_of_a_friend.name AS fofName

or

match (a)-[r1:friend]->(b)-[r1:friend]->(c)
where not(a-[:friend]->(c))

or

g.V(123).out('friend").as('1-hopf').out('friend').except('1-hopf').dedup()

or

$myfriend = go from $me over friend yield distinct friend._dst As myfriend;
$fof = go from $-.myfriend over friend  yield distinct friend._dst AS fof;
$fof minus $myfriend;

2-hop is simple to think and explain in all the above languages.
But what's the behaviour about 4-hops friends, and the duplicated intermedia nodes?
IMO, I can't be a fof or fofofof for me.

[Shylock-Hg]

If in a pure graph relationship query(not application specific). I think Q1 or Q2 is both yes.
But in specific case, if application require to remove the cyclic result , maybe we could provide a grammar to do it.

[whitewum]

Another major problem is the returning size.
for some graph computing scenario, it fetches 2 to 4 hops neighbors.
As the returning latency is critical, if you consider "I am also a friend of friend", the returning size will be very very huge.
And the latency is not acceptable in the ad hoc query.